Automatic Image Segmentation Using Edge and Marker-Controlled Watershed Transformation

computer vision. The role of image segmentation is to decompose an image into parts that are meaningful with respect to the particular applications. Subsequent methods for image description, recognition, image visualization, image compression, highly depend on the segmentation results obtained from previous stage. Therefore, we propose an automatic image segmentation method by combining an edge detection technique with modified mask as a preprocessing method and Marker Controlled Watershed Transformation (MCWT) for final image segmentation. In this paper, watershed algorithm with modified Laplacian of Gaussian (LoG) edge detector is used to detect the gradient image of input image and produce the image which is less sensitive to noise. In order to get final image with less over segmentation, it is helped by MCWT. Therefore, our proposed method has also been observed a satisfactory segmentation image with better edges and with less over segmentation.